weight1.txt

简单易懂的elaman离线学习程序

本程序为采用简化的BPTT算法离线学习的Elman网络：
每个节拍离线训练样本个数：344
校验个数：300
每个节拍最大训练次数：5000
输入层神经元个数：3
隐层层神经元个数：6
状态层神经元个数：6
输出层神经元个数：1
前后学习速率分别是：0.4	0.4	0.8
冲量系数:0.18
记忆系数:0.01

调整前权值：
0.0137867	0.0013657	-0.0284967	
0.247559	0.154706	0.0933332	
-0.0507752	0.15994	0.0615329	
-0.10107	0.0802561	0.245422	
0.0031663	0.0784249	0.232727	
0.102977	-0.177427	-0.0667669	

-0.238723	0.141507	0.12431	-0.0941267	-0.179823	0.203246	
0.216933	-0.18498	-0.0297632	-0.194792	0.200987	0.154187	
0.0954085	-0.0479827	-0.156919	0.0319605	-0.105495	0.223174	
0.126766	0.0820261	-0.0360805	0.0876873	0.120678	-0.0791116	
0.23645	0.149655	0.0548341	-0.0181661	-0.207457	-0.222625	
0.245499	-0.197386	-0.190947	-0.180509	-0.240753	0.0958205	

0.161313	-0.138836	-0.0429014	0.179151	-0.0557955	0.223418	

调整前阈值：
-0.19113	0.0944929	-0.0721686	0.158704	0.118923	-0.00653859	

0.213317	

调整后权值：
0.125467	-0.0642344	-1.02239	
-0.277795	-0.242303	-0.0772816	
-0.607547	-0.320084	-0.119338	
0.509645	0.469408	-0.56933	
-0.663732	-0.366971	0.422406	
0.536737	0.0423257	-0.968948	


0.364628	-0.605589	-0.627333	-0.379299	-1.98843	1.07864	
-0.762456	-0.441706	-0.323197	-0.983436	0.307895	-0.914727	
-0.861354	-0.286888	-0.418987	-0.706186	0.0569913	-0.856693	
2.57703	-0.735163	-0.833691	1.20373	-2.89758	3.08285	
-1.23909	0.219699	0.0892942	-0.602067	0.928484	-1.95581	
1.96868	-1.0903	-1.07372	0.238435	-2.97241	2.34764	


1.05108	-0.443854	-0.365479	2.01419	-1.05189	1.82217	

调整后阈值：
1.1023	0.92703	0.798563	1.78498	0.687485	1.10419	

0.247488